Results

Overview

Our results encompass four major areas of investigation: the engineered function of our bacterial system, fluorescent protein expression in the gut environment, transwell system validation, and L-DOPA production measurements. Select a section below to explore our detailed results.

Engineered Function

FPs in the Gut

Transwell System

L-DOPA Production

Engineered Function

Introduction

Background

GGA and the pJUMP system

Golden Gate assembly (GGA) is a DNA assembly method with the advantage of ligating up to 9 fragments of interest in a plasmid at a time.¹ GGA relies on the use of IIS restriction endonucleases, which have the property of recognizing asymmetric sequences and cleaving outside of these sites.² In synthetic biology, the ability to assemble genetic elements in predefined configurations is crucial in order to engineer organisms towards novel phenotypes of interest. Because of this, modular cloning (MoClo) stands out as a promising system to assemble functional plasmids efficiently, where hierarchical cloning can be performed to ligate sequences in a desired order.³

The joint universal modular plasmids (JUMP) system is a great example of applied modular cloning. This system introduces a range of plasmid vectors containing a wide selection of origins of replication (for different copy numbers, host ranges, etc.) and selection markers (e.g. ampicillin or kanamycin). Most importantly, the JUMP plasmids employ main cloning sites for conventional cloning followed by upstream/downstream sites meant for further vector modifications. The functionality of these sites is based on IIS endonucleases that cleave with specificity (e.g. BsaI in the main site or BsmBI in the downstream site), and make modular cloning possible. More specifically, by designing DNA sequences flanked with restriction digestion sites that are cleaved towards specific overhangs, it is possible to assemble multiple fragments in a desired order within the JUMP vectors. Based on the functionality of the plasmids and the parts they contain, the vectors can be classified as either level 0 (basic parts such as RBS, terminators, etc.), level 1 (single transcriptional units) or level 2 (multiple transcriptional units).⁴

Bacterial hosts

In synthetic biology, one of the most utilized model organisms and powerhouses for DNA amplification is Escherichia coli (E. coli). Particularly in the context of DNA cloning, E. coli DH5α stands out for its low pathogenicity and high potential in amplifying plasmid vectors.⁵ This makes E. coli DH5α an ideal candidate for molecular cloning work: transforming the cells with the GGA-assembled DNA vectors towards further analysis (mini preps, restriction digestions, colony PCRs, Sanger sequencing, etc.). A member of the same bacterial phylum (Gammaproteobacteria) is Pseudomonas alcaligenes (P. alcaligenes). This bacterium is commensal to the zebrafish gut and represents a useful target for transformations towards studying colonization.⁶

Kill switches

When facilitating the interaction of living organisms and GMOs, the design of kill switch systems is crucial to ensure control over the colonization and phenotype changes linked to the bacteria. The pBAD promoter is a highly studied genetic element, native to the E. coli araBAD (arabinose) operon and the regulatory gene araC. The distinguishing trait of this promoter is that it can be induced by the AraC protein in the presence of the arabinose carbohydrate.⁷ Therefore, the promoter can be assembled to regulate the production of mazF, a highly conserved prokaryotic toxin which carries a native role in growth regulation, due to its activity as an endoribonuclease which cleaves RNA and limits cell proliferation.⁸ The pBAD-mazF arabinose-inducible kill switch system has been described before within the iGEM competition and is part of the official iGEM registry of biological parts (BBa_K3036005).⁹

Quorum sensing is an intercellular communication mechanism in bacteria, regulating cell populations within a given space. The core of this mechanism is the LuxR-LuxI circuit, in which the latter produces a homoserine lactone (AHL), which diffuses passively out of the cell. While at low density the system does not work, a high density of bacteria leads to accumulation of AHL both intra- and extracellularly. When reaching a threshold, the lactone can work as a cofactor for the dimerization of the LuxR protein, the LuxR-AHL complex can then bind and regulate quorum-sensing-related promoters (e.g. pLux).¹⁰ In the context of kill switches, the mazF toxin can be placed under the regulation of the pLux promoter and will be expressed in highly concentrated bacterial populations. In order to finetune parameters of the desired population density, LuxR/LuxI can be expressed under different Anderson promoters, characterized by a stable and varied range of strengths.¹¹ This population control kill switch was previously described by an iGEM team (iGEM Squirrel Guangzhou 2024).¹²

Aim

The main aim of this research line was to express genes of interest in both E. coli and P. alcaligenes, by assembling pJUMP vectors with gene blocks in various configurations, and transforming them. The gene blocks we worked with were either fuGFP variants (tagged with peptides of uptake, secretion, cleavage, etc.), mCherry, TPL (tyrosinase enzyme involved in the metabolic production of L-DOPA) or kill switch elements (pBAD+mazF). We did not continue working with the second kill switch (pLux+mazF), since it encoded an incomplete circuit (see Engineering cycles). For more details on the gene blocks we utilized, please check out the Parts page of the wiki.

Following the assembly and validation of the constructs, we transformed into the host bacteria and visualized fluorescent protein production, accumulation and secretion under the Airy Scan confocal microscope. We then set out to assess the functionality of the L-arabinose-induced kill switch circuit we designed. We followed two approaches: running an overnight plate reading experiment and checking for growth hindering, followed up by plating cells on a serial dilution of the inducer and doing a growth assay.

Lastly, moving towards a functional analysis of the fluorescent constructs, we ran a plate reader experiment, measuring fluorescence in the supernatant/pellet of the cells as a means to indicate whether protein secretion takes place or not.

Validating and assessing the constructs

Prior to sending the assembled constructs for sequencing, a crucial step in preselecting promising transformed colonies was performing restriction digestion on the miniprepped plasmid DNA. For this, we utilized the restriction enzyme NotI which has recognition sites on the BioBrick prefix and suffix regions, right outside of the main cloning site in pJUMP24-1a.

Figure 1A showcases the insert sizes for a selection of the level 1 (single transcriptional units) constructs we generated and eventually validated via Sanger sequencing. Here, the expectation was to observe a 3514 bp band originating from the cut backbone, followed by bands of specific sizes based on the corresponding inserts: 895 bp (geneblock 6), 1546 bp (geneblock 7), 889 bp (geneblock 8). Since the L-arabinose kill switch (KS) circuit elements were assembled in the downstream site of the plasmid, we expected a 3977 bp band (backbone with the KS sequence) and an 885 bp one from the mother plasmid's reporter. As the gel electrophoresis image figure demonstrates, the assembly of the constructs turned out to be successful. Besides the level 1 constructs highlighted in Figure 1A, we later succeeded in validating a few more fuGFP gene blocks which were directly assembled with the mCherry block towards obtaining level 2 plasmids.

The level 2 constructs (mCherry assembled with the fuGFP variants/kill switch) were also validated by NotI restriction digestion and Sanger sequencing (Fig. 1B). Here, we expected the same cut backbone sizes, but bigger insert sizes for the validated assemblies: 1953 bp (geneblock 2), 2022 bp (geneblock 4), 1926 bp (geneblock 5), 1731 bp (geneblock 6), and 2474 bp (geneblock 7). The mCherry/KS construct was expected to yield bands of 3977 bp and 885 bp. Figure 1B showcases the band sizes for the restriction digestion products of the level 2 constructs, proving that the assemblies were successful.

Here, we highlight that the mCherry/TPL construct (geneblocks 8+7) was found to have a frameshift in the TPL gene, about 70% through the protein sequence. Since the active site of the tyrosinase is around its C-terminal, it was likely for its function to be impaired. However, due to time constraints, we decided to take it for further analysis, include it in Figure 1B and check for interesting phenotypes from this damaged L-DOPA synthesizing construct (see the L-DOPA Production Results subsection).

Figure 1: Sizes of the DNA GGA constructs. DNA agarose gel electrophoresis results of the constructs, digested with NotI, run on a 1% agarose gel. A) Band sizes of the level 1 constructs. B) Band sizes of the level 2 DNA assemblies.

Following the validation of the constructs, we transformed them into both E. coli and P. alcaligenes. Figure 2 displays Airy Scan confocal microscopy images of both bacteria transformed with a selection of the constructs. The figure shows that the bacteria give the expected phenotype, showing either red, green or both red/green fluorescence, depending on the construct. The bacteria co-expressing both fluorophores show that there are clear differences in terms of expression strength of each protein in the cells, both within and between sample groups. For example, in E. coli, geneblocks 5+8 seem to have weaker fluorescence (both channels) compared to geneblocks 4+8. This relates, of course, to the inherent biological variation seen in such experiments, and the different cell energy cost of producing various fluorophore variants. Moreso, even though the cells (both part of the Gammaproteobacteria class) have very similar, rod-like shapes, P. alcaligenes shows a tendency of clumping cells together. This is also observed in liquid cultures of the bacterium and expectedly so, since Pseudomonas bacteria are among the most prominent in medical device-associated biofilms.¹³

Figure 2: Phenotypes in bacteria transformed with the constructs of interest. Airy Scan confocal microscopy results, showcasing the (co-)expression of fluorescent proteins in E. coli and P. alcaligenes.

Testing the functionality of the L-arabinose kill switch

The implementation of a kill switch was an important part of the project, due to the potential in patient/user autonomy and safety. We used the L-arabinose-pBAD kill switch system, described in the iGEM Registry of Standard Biological Parts.¹⁴ The complete circuit was reported to contain the pBAD promoter and the mazF riboendonuclease toxin. We checked whether addition of L-arabinose would induce cell death by recording overnight growth curves of E. coli and P. alcaligenes transformed with mCherry/KS-encoding vectors in LB medium. The comparison was done with mCherry-encoding controls. We followed up with growth assays on agar plates supplemented with different concentrations of L-arabinose.

As Figure 3 displays, growth of neither of the two bacterial strains seemed to be affected upon addition of L-arabinose in the medium. More specifically, all four growth curves show lag and exponential phases of cellular proliferation at similar slopes. One potential artefact of this experiment as an indication for KS activity lies within the nature of its toxin, which cleaves RNA but does not lead to direct cell lysis (decline in the growth curve).¹⁴ As the plate reader records OD600 values, it will not differentiate between alive/dead cells. Either way, if the toxin were to be expressed, growth was expected to be significantly hindered still. For a more in depth analysis of the growth curves across conditions, along corresponding models and implications, check out our Modelling page.

Figure 3: Bacterial growth curves and kill switch functionality. OD600 values recorded overnight for bacteria expressing mCherry and mCherry/kill-switch, grown with or without the pBAD/KS inducer, L-arabinose. A) E. coli growth curves. B) P. alcaligenes growth curves.

We moved on to perform a growth assay with mCherry and mCherry/KS bacteria on LB plates (a half for each strain), supplemented with L-arabinose in different concentrations (0.16 µM to 500 µM, 5-fold steps). The results of the growth assays were recorded in the [supplementary file S1]. In P. alcaligenes, all plates displayed a smear of colonies indicating overgrowth, independent of the supplementation with L-arabinose. However, the E. coli plates yielded individual red colonies on the mCherry (control) sides, and, once again, mCherry/KS smears for the mutants. This result turned out to be inconclusive and suggested that the trend we expected in this experiment (no growth on the mCherry/KS plate halves, only for mCherry) was not met. The explanation we found for this lies within the fact that glucose (LB medium's carbon source) leads to catabolite repression. The phenomenon of carbon catabolite repression implies that catabolism of alternative carbon sources (including L-arabinose) is repressed in the presence of glucose, due to its interactions with cyclic AMP (cAMP) and its receptor protein CRP.¹⁵

As indicated by literature and past iGEM research, this finding called for the use of alternative carbon sources in the growth medium towards studying the functionality of the kill switch: glycerol.^16,17 Therefore, we repeated the growth assay on 0.2% glycerol-supplemented minimal medium agar plates with L-arabinose added in the same concentrations as before. For this experiment we also diluted all cultures to an OD600 of 0.3 and included controls: agar plates with no arabinose supplementation. The findings ([supplementary file S1]) turned out not to match our expectations this time either. While the E. coli plates did not grow any colonies, P. alcaligenes overgrew on both halves of the plates as smears but associated visible colonies of all mCherry controls. To some extent, this suggests that the kill switch does have an effect on the growth dynamics of bacteria. Therefore, while P. alcaligenes lacks the native transcription factor of the pBAD promoter (which is native to E. coli), leaky (basal) expression of the mazF could be an explanation for this observation.

Assessing the functionality of fuGFP secretion

At this stage we were also particularly interested in whether the geneblocks responsible for fuGFP secretion (HlyA-tagged, see Parts for more information), did indeed secrete green fluorescent protein. Following the overnight growth experiment described above (see more details in Modelling), we spun down the content of the plate wells with E. coli cultures we were interested in – cytosolic GFP (geneblock 6), cytosolic mCherry (geneblock 8) and mCherry/secreted GFP (geneblocks 8+5, the only secretion fuGFP construct included in the experiment). For each strain, growth had been recorded in 3 arabinose-dependent conditions, all in duplicate (6 data points/strain). The presence of L-arabinose was not expected to interfere with fluorophore secretion in the selected samples, especially since they were grown in LB medium containing glucose (repressing the catabolism of other simple carbon sources).

Fluorescence under the GFP and mCherry excitation ranges was measured in both the supernatant and resuspended pellet of each sample and the ratios were computed (Fig. 4). For GFP, the measurements (Fig. 4A) indicated significantly higher detection of green fluorescence in geneblock 6 compared to the other two, both in the supernatant and pellet. However, computing the lysate:pellet fluorescence ratios shows that the mean ratio of the secreted GFP sample is ~1.4, while cytosolic GFP/mCherry reach up to ~1.2 and ~1.1, respectively. While not statistically significant, these results indicate that the construct does indeed secrete fuGFP.

A significant factor to consider in this experiment is that the bacteria were still in the log phase at the timepoint of the spin down, which could explain fluorophore secretion-like processes. Moreso, while unlikely at this growth stage, lysis could explain the outliers with ratios of >1.7 observed in the cytosolic GFP samples. While this could also be the case for some of the secreted GFP samples taken into consideration, here all data points are >1 and show less variance compared to the cytosolic GFP, therefore we believe it's quite unlikely for this to apply to the entire sample population, if to any data point.

The fluorescence of mCherry under its excitation range is much higher in both states than the values recorded for the other two samples (Fig. 4B). Interestingly, the supernatant to pellet ratio computed from the mCherry sample turned out to be of ~1.1, which indicates approximately even distribution of red fluorescence intra- and extracellularly. An explanation for this could be, once again, a form of secretion specific to the log phase of cell proliferation. Here, we also record high fluorescence ratios for the GFP and mCherry/GFP-HlyA constructs, but this is not a biologically relevant comparison since the raw fluorescence values were much lower than those of mCherry. An interesting observation here is that the fluorophore co-expressing strain yields low fluorescence under both excitation spectra (despite giving the expected phenotype, which still shows weak fluorescence in both red/green channels, as discussed above for Fig. 2).

Figure 4: Fluorescence localization across the DNA constructs. Fluorescence values recorded for the supernatant and pellet of the bacterial strains, followed by computed ratios. Two-way ANOVA and post-hoc Bonferroni tests were performed for the comparisons between supernatant/pellet of the samples. For the ratio comparisons, Kruskal-Wallis/One-way ANOVA with Dunn's/Tukey comparison tests were used (p ≤ 0.05 *, p ≤ 0.01 **, p ≤ 0.001***, p ≤ 0.0001****).

Conclusion and Outlooks

Our initial plan was to assemble a number of 8 single-transcriptional unit constructs encoding for various functions (fluorescence, L-DOPA metabolism and induced bacterial toxicity), and then integrate them in more complex assemblies. Overall, we succeeded in generating the majority of the assemblies we envisioned for our project, the only exceptions being geneblocks 1 (fuGFP-FLAG-furin-Tat-LK15-HlyA) and 3 (fuGFP-FLAG-furin-Tat-LK15). As discussed in the Parts section, there is experimental indication that these geneblocks could have a toxic phenotype, leading to the natural selection of mutant variants with frameshifts in the plasmids of interest. Interestingly, out of our pool of mutants, these blocks share the furin-Tat sequence exclusively, which could be an interesting future study point. Of course, there is also a possibility of having received incorrectly-synthesized DNA. Sequencing of the DNA blocks we received could potentially reveal this, and also whether they are polyclonal or not. Following the transformation of the constructs in bacteria, we successfully assessed and validated the expected phenotypes in terms of (co-)fluorescence in the red/green channels of the confocal microscope.

As an important part of our project, we attempted to validate the functionality of the L-arabinose inducible kill switch using distinct methods. Despite not following the expected trends for a functioning kill switch, the data we obtained suggests that the transformed vector carries a particular functionality, to some extent. We hypothesized that a basal (leaky) level of pBAD expression could perhaps explain the limited effect of the system. However, most importantly, the inconclusive data raised interesting discussion points and ideas towards further feasible optimization for better outcomes. Here, we mention once again the utilization of glycerol as a valuable carbon source alternative, since it is not associated with inhibiting the catabolism of L-arabinose the way glucose is.¹⁶ Further, the integration of the araC regulator, even in E. coli (where it natively originates from), in the kill switch circuit is expected to be a reasonable addition to the current system. Since the modular pJUMP system was utilized in the construction of these vectors, it could easily be achievable to introduce the gene of interest in the upstream side of the already existing plasmids, thus sticking to the principles of modular cloning. It is with great interest that we suggest these next optimization steps as a way to further develop a functional kill switch for engineered bacteria, that can potentially be utilized/tested in the therapeutic context of the project.

The assessment of fluorophore secretion represented the last experimental stage of this research line, preceding the in vivo zebrafish larvae testing of the constructs. The current method gave insightful results, which suggest that the tested geneblock responsible for green fluorescent protein secretion fulfils its function. Moreso, the experimental findings highlighted that the fluorescent proteins produced by the engineered cells could likely be found extracellularly during exponential growth, which is interesting in the context of the zebrafish experiments following this section. In any case, it is worth mentioning that the current experiments do not suffice to undoubtedly claim efficient secretion of the HlyA-tagged fluorophores, as many factors are still to be considered for further analysis (i.e. LB autofluorescence, biological variance, number of replicates, etc.). If it weren't for time limitations, additional experiments could have also been performed to support the suggestions of the current data. This could include, for example, Western blots or ELISAs against the FLAG tag (present on all the proteins) to localize and quantify fluorophores. Either way, the assemblies and functionality tests described above set the foundation for insightful future experiments and opened the way for the next research lines of GutFeelings.

FPs in the Gut

Introduction

Background

Zebrafish as a model organism

Zebrafish (Danio rerio) stands out as an extensively used vertebrate model organism in the field of cell biology. In the past, it has been dominantly studied as a model for infectious systemic diseases, due to its small size, fertility and optical transparency in early developmental stages.^18,19 Relatively recent research placed the zebrafish as a key model organism in the study of the gastrointestinal (GI) tract and the gut microbiota, due to favourable comparisons to the mammalian gastrointestinal system. More specifically, it has been found that the zebrafish GI tract carries many similarities to the human one, both developmentally and physiologically (~70% of human disease genes carry orthologues in D. rerio).²⁰ Because of this, it has been possible to use zebrafish as a model to study key processes related to intestinal misfunctions, ranging from the impact of pollutants on the gut to genomics studies on irritable bowel syndrome (IBD) risk genes.²¹

Organizational and functional aspects of the zebrafish gut

Due to the high organizational and functional homology between the human and zebrafish intestines, it is no surprise that the primary role of the latter is the digestion and absorption of nutrients, followed by the elimination of waste products. Structurally, the zebrafish gut consists of three segments, each with their own functional role: the anterior intestine, the middle intestine and the posterior intestine.²¹ In the anterior intestinal bulb, there is the highest density of epithelial enteroendocrine cells.²⁰ These cells have an important role in the sensing of nutrients or other stimuli and sending signals to the nervous system, thus leading to appropriate responses (e.g. secretion of enzymes).^22,23 It is also likely for this first region to have a role in bile salt recovery. Further, the middle intestine carries a high density of goblet cells whose primary function is to secrete mucin towards the formation of a protective layer of mucus on the epithelium. The mucus layer is crucial in health as it ensures both physical support and means for chemical signaling with the gut microbiota. Commensal microbes have, then, the property of being associated with the mucus within the lumen.²⁴ In the region of the anterior-middle (predominantly middle) gut, nutrient uptake and absorption take place via enterocytes. Lastly, similarly to the human colon, the posterior region of the gut carries a role in absorbing ions and water, before waste secretion.²¹

As expected, there are also differences between the zebrafish and human intestines. In contrast to mammals, the zebrafish gut has a simpler organization which lacks a structure with homology to a stomach. Therefore, while the human stomach can reach pH values as low as 1.4, the zebrafish do not acidify their intestines. However, this does not come up with a cost for the immunity/gut function of the fish (as it would for mammals), as they express a wide range of digestive enzymes with various roles in digestion.²¹ Another big difference between the organisms under discussion relates to the components of the gut microbiota. More specifically, while there are plenty of shared bacterial divisions, the phyla of the dominant microbes differ (in fish, proteobacteria, firmicutes and fusobacteria dominate, while firmicutes, bacteroidetes and actinobacteria are more prevalent in humans).²²

Transgenic zebrafish lines used in this study

To visualize the fluorescent bacteria in the zebrafish gut, it is important to be able to localize them. To achieve this, we used two transgenic zebrafish lines that fluorescently label the vasculature. This allows for real-time imaging and orienting of bacteria spatially with respect to the vascular architecture. The first is fli1a:GFP. In this line, the promoter of the fli1 gene drives expression of GFP in all endothelial cells, marking the vasculature green.²⁵ The second is kdrl:mCherry. The kdrl promoter, an ortholog of mammalian VEGFR2, is active in vascular endothelial precursors and cells, labelling the vasculature in red.²⁶ Occasionally, larvae can be kdrl-negative, meaning they do not exhibit red fluorescence in their vasculature despite carrying the transgene.

Aim

Following the validation and assessment of our constructs in vitro, we were interested in studying the interaction of the microbes with the zebrafish larvae. More specifically, we set out to inoculate the fish with bacteria to determine whether they can colonize the gut or not. Moreso, we wanted to see where the bacteria resided in the gut and if the different constructs caused any changes in the dynamics (uptake, motility, etc.) of the fluorescent microbes. The last question of interest for this research line was whether the arabinose-induced kill switch can work in vivo and show changes in gut colonization over time. To answer these questions, we used live microscopy imaging of the zebrafish larvae inoculated with bacteria.

Results

Engineered bacteria stably colonize the zebrafish larval gut

To first establish that the P. alcaligenes strains carrying our constructs are able to colonize the zebrafish gut, we introduced bacteria carrying a level 1 construct encoding cytosolic mCherry (geneblock 8). Germ-free zebrafish larvae were exposed to this strain, and colonization was assessed by live fluorescence microscopy.Control larvae that were maintained germ-free did show low levels of background autofluorescence in the gut (Fig. 5A). By contrast, larvae fed with the mCherry-expressing bacteria exhibited stronger and spatially distinct fluorescence within the gut, indicating the colonization of the engineered bacteria (Fig. 5B). Live imaging further revealed that the fluorescent bacteria were motile within the gut environment, including the anterior gut and middle gut. Research has suggested that association of bacteria with the mucus layer in these regions, indicates stable colonization.²⁴

These results demonstrate that engineered bacterial strains carrying synthetic constructs are capable of stably colonizing the zebrafish larval gut, and that cytosolic mCherry fluorescence provides a reliable marker for tracking bacterial localization and viability in vivo.

Figure 5: Colonization of engineered bacteria in the zebrafish larval gut. Panels were stitched from several individual images to allow a larger region to be visualized. A) Tilescan images (brightfield, GFP and mCherry) of germ-free Tg(Fli1:GFP) embryo (endothelium specific expression). B) Tilescan images (brightfield, GFP and mCherry) of germ-free Tg(Fli1:GFP) embryo, inoculated with P. alcaligenes expressing cytosolic mCherry.

Video 1. Anterior gut Fli:GFP + GB8 (mCherry)

Video 2. Middle gut Fli:GFP + GB8 (mCherry)

Video 3. Posterior gut Fli:GFP + GB8 (mCherry)

Engineered bacteria colonize multiple areas of the zebrafish larval gut, including those where nutrient uptake is happening

To determine the spatial distribution of bacterial colonization in the zebrafish gut, we next examined different anatomical regions of the gut: anterior, middle, and posterior. Again, we introduced the zebrafish larvae to bacteria carrying a level 1 construct encoding cytosolic mCherry (geneblock 8). Then, fluorescence imaging was performed to assess bacterial localization.

An overview image of the larvae (Fig. 6A) illustrates the overall anatomy and positions of the three gut regions analyzed (boxes 1, 2 and 3). A schematic (Fig. 6B) depicts the anterior, middle, and posterior regions. Figure 6B highlights the presence of enteroendocrine cells in the anterior gut, which play an important role in the sensing of nutrients and other stimuli and sending signals to the nervous system, and goblet cells in the middle gut, which play a crucial role in forming the protective layer of mucus on the epithelium. Consistent with this organization, in the region of the anterior-middle (predominantly middle) gut, nutrient uptake and absorption takes place via enterocytes.

Fluorescence microscopy revealed the presence of motile mCherry-expressing bacteria throughout the gut, specifically the anterior, middle, and posterior regions (Fig. 6C). In all regions, fluorescent signal was observed within the intestinal lumen, which indicates bacterial colonization.

In the middle gut, where nutrient uptake and absorption occurs, fluorescent signal was also observed in the epithelial cell layer. In previous platereader results, we confirmed the presence of soluble mCherry, these experiments were run within 24-hours, and assessment of OD600 showed bacteria were still in log phase, meaning this soluble mCherry is unlikely due to lysed bacteria, rather a form of secretion is happening.

These observations indicate that engineered bacteria are capable of colonizing multiple regions of the zebrafish gut, including the anterior-middle gut where uptake occurs and goblet cells are present. This supports the potential of using these bacterial strains for local production and delivery of compounds in areas of the gut associated with nutrient absorption.

Figure 6: Colonization of engineered bacteria in distinct regions of the zebrafish larval gut. A) Tilescan images (brightfield + GFP) of germ-free Tg(Fli1:GFP) embryo (endothelium specific expression), inoculated with P. alcaligenes expressing cytosolic mCherry. Boxes 1, 2 and 3 indicate different regions of the gut shown in 6C. Panels were stitched from several individual images to allow a larger region to be visualized. B) Schematic image indicating the anterior, middle, and posterior region of the zebrafish larval gut, highlighting the presence of enteroendocrine cells and goblet cells. C) Confocal images showing bacterial colonization in the anterior, middle, and posterior region of the zebrafish larval gut, localization as indicated by the boxes shown in figure 6A.

Engineered bacteria show differences in uptake

Following the assessment of our constructs in vitro and validating colonization of the gut by our bacteria, we were interested in assessing if our different constructs caused any changes in the dynamics of our fluorescent proteins. For this, we compared three level-2 constructs and a level-1 construct, namely:

Geneblock 8, which expresses cytosolic mCherry;
Geneblock 8 + 2, which expresses cytosolic mCherry and fuGFP with a secretion tag (HylA) and furin;
Geneblock 8 + 4, which expresses cytosolic mCherry and fuGFP with a secretion tag (HylA) and uptake tag (Tat-LK15);
Geneblock 8 + 5, which expresses cytosolic mCherry and fuGFP with a secretion tag (HylA).

Larvae fed with a level-2 construct again exhibited strong fluorescence within the gut (Fig. 7A). In all conditions, GFP signal was seen along the gut overlapping with mCherry signal, indicating colonization of our dual-fluorescent bacteria (Fig. 7A, 7B).

One of our goals was to evaluate the differences in secretion and uptake of GFP between the different constructs in vivo. In vitro, we confirmed the presence of soluble protein for both mCherry and fuGFP (see the Engineered Function Results subsection), where assessment of OD600 showed bacteria were still in log phase, indicating some form of secretion is happening. However, in vivo we were not able to detect secretion and / or soluble protein for both mCherry and fuGFP. This could be explained by background fluorescence, or that the fluorescent proteins are quickly taken up or degraded.

Interestingly, in contrast to mCherry uptake, no uptake of GFP signal by endothelial cells was detected in any of the conditions. One explanation is that fuGFP fluorescence is quenched following internalization into acidic compartments, such as endosomes. Many GFP variants lose detectable fluorescence at mildly acidic pH, while mCherry fluorescence is more stable.²⁷ This means fuGFP uptake could still be taking place, but we are unable to visualize this.

When larvae were colonized with geneblock 8 + 2, we observed uptake of mCherry into the intestinal epithelial cell layer (Fig. 7B), similar to the uptake seen with the level-1 mCherry strain (Fig. 6C). Surprisingly, geneblock 8 + 4 did not show detectable uptake of mCherry (Fig. 7B). Since geneblock 8 + 2, geneblock 8 + 5 and level-1 construct geneblock 8 show similar uptake in mCherry, we assume that the furin cleavage module plays no role in the observed difference between geneblock 8 + 2 and geneblock 8 + 4. This leads us to think that the presence of Tat-LK15, a cell penetrating peptide (CPP) or 'uptake tag', on fuGFP could influence mCherry uptake dynamics. CPP-mediated internalization is known to involve receptor- and endocytosis-dependent pathways that can become saturated or can be competed with, such that the presence of an additional uptake-tagged cargo (fuGFP-Tat-LK15) could shift the uptake dynamics of co-secreted proteins (mCherry) that share receptors or endocytotic machinery.²⁸ Moreover, studies have shown that the efficiency and route of Tat-mediated uptake can vary depending on the attached cargo and its abundance.²⁹ It is therefore plausible that co-secretion of an additional uptake-tagged protein (fuGFP-Tat-LK15) may partially occupy shared uptake machinery or receptors, thereby shifting the uptake dynamics of mCherry.

We realize we have to interpret the absence of detectable GFP uptake and differences in mCherry uptake carefully. Due to time restraints, we only have one reproduction of each of these conditions, thus more research is needed to elucidate the uptake mechanisms of our co-expressed fluorescent proteins, and how the presence of an uptake tag, such as Tat-LK15, influences this. To verify if uptake of GFP is happening, experiments could be repeated, but with a pH-stable fluorophore. One example of this could be the acid-tolerant monomeric GFP 'Gamillus' from the jellyfish Olindias formosa, developed by Shinoda et al. (2018).²⁷ Experiments with lysosomal and endocytosis inhibitors might elucidate on the mechanisms by which uptake is happening. One example could be to use the vacuolar H(+)-ATPase inhibitor Bafilomycin A (BafA). BafA increases endosomal pH and thereby decreases lysosomal degradation. A study by Verweij et al. (2019) showed that this leads to accumulation of exosomes in endothelial cells, suggesting that lysosomal degradation limits detectable intracellular signals.³⁰ Applying BafA in our system could reduce degradation of secreted fluorescent proteins, and possibly cause quenched GFP to remain visible, because of the increase in pH. This could provide more insight into the intracellular fate of our fluorescent protein after uptake.

Figure 7: Zebrafish larvae colonization of bacteria carrying level-2 constructs. A) Tilescan images (brightfield, GFP and mCherry) of germ-free Tg(kdrl:mCherry) larvae, inoculated with P. alcaligenes expressing cytosolic mCherry. B) Confocal images showing bacterial colonization for control, geneblock 8 + 2 and geneblock 8 + 4 in the anterior and middle region of the zebrafish larval gut.

In vivo activation of an arabinose-induced kill switch in gut colonizing bacteria

Lastly, we evaluated whether our arabinose-inducible kill switch could function in vivo. Germ-free zebrafish were colonized with engineered P. alcaligenes containing the kill switch and cytosolic mCherry labeling (Geneblock 8 + KS1). Fluorescent imaging was performed prior to and 16 hours after the addition of L-arabinose (Fig. 8). The last question of interest for this research line was whether the arabinose-induced kill switch shows changes in bacterial gut colonization over time. In vitro, proof that our killswitch was successful remained inconclusive.

In the pre-induction images (pre L-arabinose), mCherry fluorescence was detectable in the gut lumen region, indicating successful colonization (Fig. 8A). Following L-arabinose addition, overall fluorescence intensity and bacterial signal seemed to decrease substantially, suggesting partial activation of the kill switch and loss of bacterial colonization (Fig. 8B). However, residual fluorescence remained and changes in fluorescence might also be explained by other factors, such as fluctuations in fluorescence in Z or over time. Due to time constraints no positive control was performed, that compared the colonization of bacteria without addition of L-arabinose over time.

These results show preliminary in vivo functionality of the kill switch, where it reduces bacterial colonization over time. However, further optimization and validation are needed to achieve complete clearance.

Figure 8: In vivo activation of an arabinose-inducible kill switch in gut colonizing bacteria. A) Tilescan images (brightfield + mCherry) of germ-free Tg(kdrl:mCherry) zebrafish larvae colonized with P. alcaligenes expressing Geneblock 8 + KS1. Images show gut colonization before (pre) and 16 hours after (post) addition of L-arabinose, the chemical inducer of the kill switch. B) Magnified images of the anterior gut from the same individuals, comparing bacterial signals before and after L-arabinose addition. The partial reduction in mCherry indicates kill switch activation, though not full clearance.

Conclusion

Our results demonstrate that our engineered bacterial strains carrying synthetic constructs are capable of stably colonizing the zebrafish larval gut, both for a level-1 construct, carrying only cytosolically expressed mCherry, and for level-2 constructs, with sequences coding for different variants of fuGFP. Moreover, our data shows that the bacteria localize throughout the anterior, middle, and posterior regions of the gut, thus including regions where nutrient uptake and absorption occur. Moreover, we could also see uptake of mCherry by endothelial cells lining the gut in these regions. This colonization of physiologically relevant areas is a good foundation for the future of engineering bacterial strains aimed at delivery of drugs in the gut environment.

One of our goals was to evaluate the differences in secretion and uptake of GFP between the different constructs in vivo. No uptake of GFP signal by endothelial cells was detected in any of the conditions. One explanation is that fuGFP fluorescence is quenched following internalization.²⁷

We saw striking results when looking at the uptake of mCherry across our different constructs, the presence Tat-LK15 on fuGFP seems to decrease mCherry uptake. This could be explained by the fact that co-secretion of an additional uptake-tagged protein (fuGFP-Tat-LK15) partially occupies shared uptake machinery or receptors, thus shifting the uptake dynamics of mCherry.^28,29

We realize we have to interpret the absence of detectable GFP uptake and differences in mCherry uptake critically. Due to time restraints, we only have one reproduction of each of these conditions, thus more reproductions are needed to make this data conclusive. Furthermore we propose several experimental set-ups to elucidate the uptake mechanisms further, such as using a pH stable GFP²⁷ or by using the H(+)-ATPase inhibitor BafA to possibly cause quenched GFP to remain visible and accumulate in endothelial cells.

The last question of interest for this research line was whether the arabinose-induced kill switch shows changes in bacterial gut colonization over time. Results show preliminary in vivo functionality of the kill switch, where it reduces bacterial colonization over time. However, the collection of more data and including stable controls would greatly help with reaching accurate conclusions. The environment of the zebrafish microbiome is complex, both biologically and chemically. Therefore, further optimization and validation experiments are needed to achieve complete clearance, both in vivo and in vitro (see the Engineered Function Results subsection). This is particularly important considering the therapeutic aspect of the project, since such a system should have a complete success rate in removing the engineered microbes from the microbiome of a patient.

Even though further research is needed to elucidate the uptake mechanisms of our engineered fluorescent proteins, we believe our results are still very promising. We have shown that our engineered bacteria can stably colonize various regions of the zebrafish larval gut, can secrete proteins here, which can be taken up by endothelial cells lining the gut. Besides, we have made efforts towards a safety-control system in the form of a killswitch. Despite challenges, our results demonstrate a promising step towards a living biopharmaceutical and a new way of drug delivery.

Transwell System

Introduction

Engineering bacteria to deliver therapeutic molecules across the intestinal barrier represents a promising approach for medicine delivery in the gut. To analyze this concept in a human-relevant setting, we used a Caco-2 Transwell-based intestinal model adapted from Floor et al. (2025). In this system, Caco-2 cells are cultured under an air liquid interface (ALI) with basolateral vasoactive intestinal peptide (VIP) supplementation. This treatment promotes the formation of a polarized epithelial monolayer covered by a dense mucus layer, closely mimicking the human intestinal lining. The model enables studies of bacterial colonization, bacterial-mucus interaction, secretion of bioactive compounds, and transport across the epithelial barrier into the basolateral compartment.

For our experiments, we focused on four bacterial strains: L-DOPA–producing Pseudomonas alcaligenes, L-DOPA–producing E. coli, GFP-producing E. coli, and a non-fluorescent E. coli negative control. While we initially considered using a human commensal bacterium such as Lactobacillus, time constraints prevented its transformation and validation before the experiment. Pseudomonas has already been used in zebrafish, which makes it interesting to compare its behaviour in the Transwell system, while E. coli represents a more established and well studied bacterial model. Even if these strains are not intended for human use, these experiments offer a baseline for comparing in the future, suitable chassis candidates and can support chassis design and biosafety assessments.

Fluorescent reporters and L-DOPA production were used to monitor bacterial behavior and functional output. Cytosolic red fluorescence (mCherry) enabled us to track bacterial localization within the mucus layer, while green fluorescence (GFP) allowed for the detection of secreted proteins and their potential uptake by Caco-2 cells.

Using this setup, we aimed to:

Evaluate the ability of engineered bacteria to colonize the mucus layer.
Quantify production and secretion of therapeutic outputs (L-DOPA or fluorescent proteins).
Measure transport of secreted molecules across the epithelial barrier into the basolateral compartment using fluorescence or HPLC-based assays.

Experiment 1: Transwell Testing with Added Compounds

Results and Discussion

Before introducing engineered bacteria or testing L-DOPA delivery, we first assessed cell viability to characterize the toxicity of our tested compounds at their respective concentrations. This step was necessary to ensure that the epithelial monolayer and mucosal layer would remain viable when exposed to the components required for L-DOPA production. We tested the basolateral addition of ascorbic acid at three concentrations (100, 250, and 500 µM), apical addition of different concentrations of our working dilution, while control wells received standard medium, and monitored epithelial integrity using TEER measurements. This approach allowed us to establish optimal working conditions for future experiments.

Table 1: Added Working Solution

Sample name	Apical Side (V_Total = 150 µL)	Basolateral Side (V_Total = 600 µL)
Control (S1)	150 µL medium	600 µL Medium
Sample 1 (S2)	150 µL working solution	150 µL working solution + 450 µL DMEM
Sample 2 (S3)	150 µL working solution	150 µL working solution + 450 µL 100 µM AA
Sample 3 (S4)	150 µL working solution	150 µL working solution + 450 µL 250 µM AA
Sample 4 (S5)	150 µL working solution	150 µL working solution + 450 µL 500 µM AA

Figure 9: Plate layout.

Table 2: TEER Results at t=0 and t=4, After a 4 Hour Incubation Period

Sample name	TEER initial (Ω·cm²)	TEER post incubation (Ω·cm²)
Control (S1)	318	346
Sample 1 (S2)	377	140
Sample 2 (S3)	303	135
Sample 3 (S4)	320	148
Sample 4 (S5)	335	164

Discussion

Our results indicate that the current working mixture is toxic to Caco-2 cells, likely due to both chemical composition and osmotic effects. This finding shows the importance of optimizing medium preparation for small-molecule production without affecting the viability of the intestinal epithelial monolayer. Moving forward, we will use a toxicity assay to establish optimal working concentrations for L-DOPA production.

Experiment 2: Toxicity Assay

Given the time-intensive and limited repeatability nature of preparing Caco-2 Transwells, we performed a preliminary toxicity assay to evaluate the compatibility of the working solution with epithelial cells. The goal was to determine safe concentrations for future experiments with L-DOPA producing bacteria. Caco-2 cells were seeded in a 96-well plate and exposed to a serial dilution of the working solution containing ammonium acetate, sodium sulfite, sodium pyruvate, and pyrocatechol. LDH release was measured as an indicator of cell viability using phenol red–free medium, with 1% Triton X-100 as a positive control and untreated cells as a negative control.

Table 3: Layout of Medium Samples Collected from the Toxicity Assay

WD = Working Dilution

Row	1	2	3	4	5	6	7
a	1x WD	1:2	1:4	1:8	1:16	1:32
b	1:64	1:128	1:256	1:512	1:1024	Positive control	Negative
c	1x WD	1:2	1:4	1:8	1:16	1:32
d	1:64	1:128	1:256	1:512	1:1024	Positive control	Negative

Table 4: LDH Release Measurements in Caco-2 Toxicity Assay

Row	1	2	3	4	5	6	7
a	3.385	3.404	3.500	3.500	3.500	2.260
b	1.813	1.359	1.322	0.882	0.782	0.669	3.500
c	3.069	3.162	3.023	2.972	3.500	2.170
d	1.364	1.560	1.029	0.818	1.032	0.925	3.500

Discussion

LDH measurements indicated that the undiluted working solution was toxic to Caco-2 cells, with cytotoxicity decreasing at higher dilutions. However, the results were inconclusive because the positive and negative controls were most likely mixed up during the assay. This mix-up means that the absolute LDH values cannot be directly interpreted, and that nothing can be deduced from the obtained results. Due to time constraints, we were unfortunately not able to re-run this experiment. Despite this uncertainty, the data suggested that only highly diluted concentrations of the working solution would be compatible with epithelial cell-viability. After discussion with supervisors, we decided that the Transwell experiment should proceed without additional precursors or supplements, relying instead on the bacteria to produce L-DOPA themselves. This decision prioritizes the integrity of the epithelial monolayer, as any residual toxicity could compromise the success of the experiment. Even if L-DOPA production is limited, this approach will still allow us to assess bacterial mucus-colonization without impairing the epithelial barrier function.

Experiment 3: Bacterial Infection

We used four bacterial strains in our Transwell system: L-DOPA–producing Pseudomonas alcaligenes gb8+7, L-DOPA–producing E. coli gb8+7, GFP-producing E. coli gb8+5, and a non-fluorescent E. coli negative control. Each strain was applied to the apical side of differentiated Caco-2 Transwells to assess bacterial mucus-colonization, localization, and the secretion of fluorescent proteins and/or L-DOPA. Due to time constraints, only four Transwells were prepared, and two were ultimately imaged.

Results

Figure 10: Magenta fluorescence indicates Pseudomonas alcaligenes strain gb8+7, green marks the mucus layer (Muc2), and blue corresponds to epithelial cell nuclei (DNA). Confocal imaging of the Transwell coculture system (40× oil immersion) illustrates the organization of the in vitro intestinal mucosal model. The central panel presents a top-view projection of the Transwell insert, while the lower and side panels show orthogonal cross-sections. The images demonstrate that strain gb8+7 adheres to and colonizes the mucus layer in the in vitro gut model.

The Caco-2 cells form a continuous epithelial layer covered by a well-defined and intact mucus layer. Distinct magenta rod-shaped Pseudomonas cells can be observed, indicating bacterial colonization at the mucus interface. The orthogonal sections further confirm that the bacteria are positioned within or just above the mucus layer. Occasional cyan signal throughout the field likely represents background fluorescence or extracellular bacterial debris.

Overall, the imaging demonstrates colonization of the mucus layer by Pseudomonas while maintaining epithelial integrity. This spatial organization mirrors the pattern reported for human commensal Lactiplantibacillus plantarum in the reference model, suggesting that the L-DOPA–producing Pseudomonas interacts primarily with the mucus rather than intestinal epithelial cells, something observed previously with more pathogenic bacteria such as Salmonella.

Figure 11: Cyan fluorescence indicates L-DOPA–producing E. coli, green marks the mucus layer (Muc2), and blue corresponds to epithelial nuclei (DNA). Confocal imaging of the Transwell coculture system (40× oil immersion) shows the epithelial layer with intact Caco-2 cells and overlying mucus. Only a few bright cyan spots are visible on the mucus surface, and faint puncta are observed within or above the mucus, near background levels. These images suggest that E. coli either colonizes the mucus at low abundance or expresses insufficient mCherry for reliable detection.

The epithelial layer displayed the characteristic morphology of the cells, with an intact mucus layer overlying the Caco-2 cells. However, the mCherry signal from the L-DOPA–producing E. coli was weak, and only limited bacterial fluorescence could be detected within the system. Very few bright spots can be seen sitting on top of the mucus layer, and faint cyan puncta were observed above or within the mucus layer; however, these signals were at near-background levels and difficult to distinguish from diffuse fluorescence.

Despite the low fluorescence intensity, the data suggest that the E. coli cells were either present at low abundance or expressed insufficient levels of mCherry for reliable visualization. There is high background information, after discussion with supervisors.

Figure 12: DAPI-stained Transwell system

Figure 12: Confocal imaging of the Caco-2 ALI-VIP Transwell system (40× oil immersion) shows the epithelial layer with DAPI-stained nuclei (blue), marking both Caco-2 cells and bacterial DNA. No mucus layer is shown. The central panel presents a top-view projection of the Transwell insert, while the lower and side panels show orthogonal cross-sections. Images were acquired at higher exposure to facilitate bacterial detection. Small bright blue spots, consistent with bacterial size, are visible across the apical surface, indicating that L-DOPA–producing E. coli successfully colonized the mucosal surface. DAPI staining provided improved visualization of bacteria compared to mCherry fluorescence.

DAPI staining provided improved visualization of the bacteria compared to mCherry fluorescence. In these images, small blue bright spots, consistent with bacterial dimensions, can be observed distributed across the apical surface of the mucus layer. As shown in the panel below, bacteria were located on top of the mucus, suggesting colonization of the mucus surface by the L-DOPA-producing E. coli.

Initially we planned to perform quantificaiton of L-DOPA production with LC-MS/MS. However, the week of measurement the machine broke. Unfortunately, due to the lack of LC-MS/MS, quantification was impossible since L-DOPA concentrations were below the limit of detection for HPLC.

Discussion

This study demonstrates the successful use of a Caco-2 transwell system that enables visualization and investigation of bacterial-mucus interactions. We demonstrate that the presence of mucus in this in-vitro system allows colonization experiments with engineered bacterial strains. The system performed as intended: the Pseudomonas bacteria were clearly visible by confocal microscopy and localized mainly on top of the mucus layer.

Our observations with Pseudomonas alcaligenes suggest, surprisingly, that this strain interacts with the mucus interface like that of the commensal bacterium Lactiplantibacillus plantarum. The bacteria remained confined to the apical side and did not invade the epithelium, consistent with the previously described spatial organization of Lactibacillus. This was an unexpected outcome since Pseudomonas alcaligenes is considered an opportunistic pathogen for humans, while human bacterial infections are not common, we still expected it to exhibit pathogenic traits within the Transwell system.³¹

However, this proof-of-concept study was limited to a short 2-hour bacterial infection period and a small number of replicates, so quantitative conclusions about bacterial adhesion or mucus penetration could not yet be drawn. The weak fluorescence signals from the E. coli strains, likely due to low expression (in our zebrafish experiments we also noticed that E. coli produces less mCherry), further highlight the need for optimized labeling strategies in future. Extending infection time courses, improving fluorescent protein expression, and including live-cell imaging could provide greater insight into bacterial dynamics and mucus interactions over time.

A major next step will be to examine how the bacterial presence influences mucus composition and intestinal barrier function. Measuring mucus-associated proteins (e.g. mucin 2), and host secreted factors after bacterial exposure would clarify how different species interact with or modulate the mucus layer. As well as, optimising the set up to facilitate proper L-DOPA production and quantification, to measure transport of secreted molecules across the epithelial barrier into the basolateral compartment. Ultimately, the use of human commensal bacteria will allow for more physiological assessment of host microbe compatibility and the ability of commensal engineered bacteria to produce drugs without harming the system.

This Transwell system offers potential as a platform to evaluate bacteria-mediated drug delivery. By maintaining both mucus integrity and epithelial viability, the model could help bridge the gap between in-vitro screening and in-vivo intestinal applications. Overall, our findings confirm that the proof of concept worked as intended and establish a reproducible framework for future studies into microbial colonization, mucus biology, and therapeutic bacterial delivery.

Experiment 4: GFP Measurement

To assess the translocation of GFP across the epithelial barrier, we collected basolateral medium from Caco-2 the Transwell that had been exposed to GFP-producing E. coli. Fluorescence was measured using the Typhoon™ biomolecular imager, specifically attempting to detect the emission signal of GFP.

Figure 13: Top 3 wells: Basolateral medium from exposed Transwell. Bottom 3 wells: Negative control, only medium. Fluorescence detection at excitation wavelength at 473 nm.

Discussion

Unfortunately, no difference in fluorescence was observed between the GFP-exposed samples and the negative controls. This was likely due to two factors: the presence of phenol red in the medium, which absorbs light at similar wavelengths as GFP, and the anticipated low levels of GFP that could diffuse across the epithelial monolayer. As a result, all measurements exceeded the range of detection, preventing any quantification or even measurement. Consequently, we were unable to determine whether any GFP crossed the epithelial barrier or even had been produced. Future experiments need to be performed in phenol red free medium.

L-DOPA Production

Introduction

Background

L-DOPA and the physiological response

The substantia nigra (SN) is a basal ganglia structure, located in the midbrain. It represents a very important dopaminergic nucleus (produces and uses dopamine as a neurotransmitter) and it has a crucial function in regulating motor movements.^32,33 In Parkinson's disease, motor symptoms stand out as a consequence of SN cell loss and damage of the corresponding nigrostriatal dopaminergic pathways, which are essential for voluntary movement.³⁴ Levodopa (L-DOPA) is a dopamine precursor, used as medication to treat patients of Parkinson's disease. Treatment with L-DOPA has been a cornerstone in the treatment of the disease ever since its introduction over 60 years ago, due to its high efficacy and favourable safety profile.³⁵ When administered orally, levodopa can turn to dopamine quickly, via a decarboxylation reaction. Therefore, in order to prevent the reaction from happening before the compound can reach cerebral tissues, levodopa is usually taken with a dopa-decarboxylase inhibitor such as carbidopa. This prevents the symptoms linked to premature dopamine accumulation (e.g. hypotension or nausea), while extending the half-life and effect of the drug.³⁴

When taken together with carbidopa, the L-DOPA treatment has been associated with disturbing some cardiovascular metrics (e.g. mean arterial pressure or cardiac contractility), with others remaining constant, including systemic vascular resistance and heart rate.³⁶ However, when the decarboxylase inhibitor is missing and dopamine can accumulate in extracerebral tissues, it is likely for the side effects mentioned above to have cardiovascular impact. For example, hypotension may indirectly lead to an increase in the heart rate, as the organ compensates for the drop in blood pressure.

The dopamine signaling circuits are conserved across the animal kingdom. Even though the zebrafish model organism lacks a direct homologue to the mammalian nervous system, its dopaminergic cells have been reported to bear roles in regulating locomotor activity similar to those of mammals/humans. Therefore, zebrafish could be employed in past research to study neurological disorders, including Parkinson's disease with L-DOPA.³⁷ Keeping this in mind, it is likely for L-DOPA/carbidopa to lead to similar trends of physiological phenotypes in zebrafish as reported for human patients.

Ex vivo bacteria-based production of L-DOPA

L-DOPA Production

Although L-DOPA (Levodopa) is widely recognized as the mainstream treatment for Parkinson's disease, it has a complicated eight-step chemical synthesis and is associated with long-term side effects. After approximately five years of use, many patients start developing dyskinesia, thought to result from fluctuating plasma drug levels that lead to unsteady drug supply to the brain.³⁸

In recent years, bacteria-based production of L-DOPA has been established as a potential replacement or enhancement of the industry standard. Following this interest, we propose achieving a biological system for sustainable and controlled L-DOPA production using gut bacteria. We propose using an engineered commensal bacterium to produce L-DOPA in situ. We are going to use Pseudomonas alcaligenes, a commensal zebrafish gut bacterium, as a microbial chassis. However, before the proof of concept studies were performed in the gut of zebrafish larvae, we wanted to achieve production outside of the gut first.

Production Design

Tyrosine phenol-lyase (TPL) is a pyridoxal phosphate dependent enzyme, and is the central enzyme of interest for L-DOPA production in the gut.³⁹

Figure 14: Reactions catalysed by the multi-purpose enzyme, Tyrosine phenol-lyase (TPL). α,β-elimination of L-tyrosine to produce phenol, pyruvate, and ammonia. β-replacement reactions, such as condensation of pyrocatechol, pyruvate, and ammonia to yield L-DOPA. Synthetic and racemization reactions with various amino acids. (Kumagai et al., 2023)

The α,β-elimination reaction, as shown in Figure 14, is reversible; thus under optimal precursor conditions (pyrocatechol, pyruvate, ammonia), we can shift the equilibrium towards β-replacement and drive L-DOPA synthesis.⁴⁰

There are two main ways to induce L-DOPA, one through TPL or by employing tyrosinase. Tyrosinase catalyzes the oxidation of L-tyrosine to L-DOPA (cresolase activity) and then to DOPAquinone (Fig. 15). However, tyrosinase readily overoxidises L-DOPA to DOPAquinone, which then polymerizes into melanin.⁴¹ Additionally, this reaction requires oxygen, which is not as readily available in all parts of the intestine. This may reduce yield and complicate the process, making the TPL pathways more attractive for controlled synthesis with limited risk of rapid overoxidation.

Figure 15: Reaction scheme of the L-DOPA production pathway via tyrosinase. (Xu et al., 2012)

Following this, we have selected a protocol based on TPL-producing bacteria to start several 100 ml cultures to produce L-DOPA ex-vivo to test our plasmid. This protocol has previously been done in E. coli, not P. alcaligenes, however literature does indicate that it has also been able to produce L-DOPA.³⁹

Aim

As part of this research line, we set out to follow 2 main distinct directions: ex vivo bacteria-based production assessment and in vivo phenotyping. Firstly, we want to elucidate whether our bacteria expressing the TPL enzyme, the enzyme responsible for metabolic synthesis of L-DOPA, can produce L-DOPA ex vivo. After this, we intended to elucidate whether our constructs expressing the TPL enzyme could be linked to changes in the physiology of the zebrafish larvae. More specifically, after inoculating the fish with our TPL expressing bacteria, we recorded time-lapse videos of the hearts and used those to extrapolate average heart rates for each condition.

Results

We tested multiple transformed bacteria strains under the conditions outlined in our ex-vivo production protocol. The following constructs were analysed:

E. coli level 1 plasmid (geneblock 7);
P. alcaligenes:
- Control (no plasmid);
- Geneblock 7 (with the reaction mix);
- Geneblock 7 (no reaction mix, just fed substrates).
Final round was E. coli and P. alcaligenes:
- P. alcaligenes geneblock 8;
- P. alcaligenes geneblocks 7+8;
- P. alcaligenes geneblock 7 (level 1);
- P. alcaligenes control;
- E. coli DH5a (control);
- E. coli geneblock 8;
- E. coli geneblocks 7+8.

All samples were analysed by HPLC under the optimised quantification method (see Protocols). No detectable L-DOPA was observed in any of the tested conditions. Chromatograms from all showed no peaks at the expected retention time, and UV spectra didn't show L-DOPA absorbance signature (Fig. 16, Fig. 17).

Figure 16: Chromatogram results. A) Chromatogram of P. alcaligenes with geneblock 7, with methyl-DOPA as the internal standard. B) Chromatogram of LB spiked with methyl-DOPA and L-DOPA. Methyl-DOPA and L-DOPA always co-elute in this system, where L-DOPA elutes after methyl-DOPA.

As a part of our HPLC quantification protocol, we prepared a 6 point standard curve with concentrations between 11.25-360.0 µg/mL, methyl-DOPA being an internal standard. Table 5 showcases the peak areas, back-calculated concentrations, and recovery rates.

Figure 17: L-DOPA standard curve. Standard curve for L-DOPA quantification, plotted from 6 data points (11.25 to 360.0 µg/mL).

Table 5: Standard Curve Results

L-DOPA conc (µg/mL)	L-DOPA AUC	Methyl-DOPA AUC	Back Calculation	% recovery
360	121849	125780	366.03	101.68
180	54604	137721	171.64	95.36
90	23182	131767	80.81	89.79
45	10616	132646	44.48	98.85
22.5	4500	133438	26.80	119.13
11.25	1793	132340	18.98	168.70

The standard curve is linear across 45-360 µg/ml, with recoveries within ±15%. However, at lower concentrations (22.5 µg/ml and lower), recoveries deviate significantly, indicating higher variability near the limit of detection of the HPLC-UV method. Therefore, our method is not reliable for such low concentrations, and LC-MS/MS will be needed to confirm trace levels of production.

L-DOPA production and effects on zebrafish larval heartbeat

We also explored the physiological impact of L-DOPA production by engineered bacteria. To this end, germ-free zebrafish larvae were colonized with P. alcaligenes carrying the geneblock 8 + 7 construct. In this design, Geneblock 7 encodes tyrosine phenol lyase (TPL) for L-DOPA biosynthesis, while Geneblock 8 provides cytosolic mCherry expression for bacterial visualization. Sequencing later revealed a frameshift within the TPL cassette (See Engineering Cycle cloning #4), suggesting that enzyme expression and L-DOPA production were likely impaired.

In mammals, L-DOPA administered without carbidopa, a dopa-decarboxylase inhibitor, is known to lower blood pressure, which in turn increases heart rate.³⁴ Because dopaminergic signaling pathways are conserved across vertebrates,³⁷ we aimed to assess whether bacterial L-DOPA production could influence zebrafish cardiac physiology. Zebrafish are particularly suited for such analysis, as their baseline larval heart rate (150–225 bpm) closely resembles that of humans.⁴²

An overview image of a kdrl-negative larva (lacking vascular mCherry signal) colonized with P. alcaligenes geneblocks 8+7, shows bacterial gut fluorescence (Fig. 18A), with magnified images confirming bacterial localization within the middle gut (Fig. 18B).

To assess potential physiological effects, we quantified cardiac rhythm from time-lapse imaging of kdrl:mCherry larvae across four different conditions: control kdrl:mCherry, control + 1 mM L-DOPA, geneblock 7 (Level 1), and geneblocks 8+7 (Level 2). Heart beat traces were extracted by measuring the mean fluorescence intensity over time within a fixed region encompassing the entire heart, ensuring comparable areas across all conditions. The recorded signal fluctuated as the heart contracted and relaxed, producing periodic intensity peaks corresponding to cardiac cycles. Although smaller ROIs produced more distinct traces, they also increased variability; hence, a standard ROI covering the whole heart was used for all samples to maintain consistency. We drew methodological inspiration from Verweij et al., (2019),⁴³ who similarly used fluorescence intensity fluctuations to monitor physiological processes, such as signaling gradients evolving over time. Control kdrl:mCherry larvae showed a clear and regular cardiac rhythm. Control + 1 mM L-DOPA larvae displayed irregular ("arrhythmic") traces, suggesting altered cardiac periodicity. Geneblock 7 (Level 1) larvae colonized with P. alcaligenes expressing TPL showed intermediate patterns between rhythmic and irregular traces. Lastly, geneblocks 8+7 (Level 2) larvae exhibited an irregular trace profile, similar but less extreme than seen in control + 1 mM L-DOPA (Fig. 18C).

To extract beats per minute (bpm), the number of peaks was quantified using the BAR → Find Peaks ImageJ plugin. The average heart rate of control kdrl:mCherry zebrafish larvae from two replicates was 225 bpm. Control kdrl:mCherry zebrafish supplied with 1mM L-DOPA resulted in a reduction in heart rate to 176 bpm. Larvae colonized with the Level 1 GB7 construct showed an increased heart rate to 278 bpm. Those colonized with the Level 2 GB8+7 strain had a heart rate of 206 bpm (Fig. 18D).

However, these values should be interpreted cautiously. The dataset consisted of single replicates per condition, and heartbeat detection was performed using non-optimized, semi-manual methods. Motion, background fluorescence, and noise likely affected the signal quality, resulting in high variability and inconclusive BPMs between conditions. Although the control group's rate was within published ranges for zebrafish larvae (150–225 bpm), 1mM L-DOPA treated and GB8+7 larvae's rates were lowered, possibly due to measurement artifacts and the dysfunctional TPL gene. The elevated heart rate observed in GB7 larvae might suggest a potential stimulatory effect associated with TPL expression; however, this is unlikely given the frameshift. More plausibly, this observation reflects signal noise rather than a genuine physiological response, considering the limited data and measurement uncertainty.

Given these limitations and the frameshift in the L-DOPA construct, these results cannot confirm whether bacterial L-DOPA production directly modulates cardiac activity. For future work, we recommend increasing the number of replicates and applying validated, potentially automated heartbeat detection methods for zebrafish larvae. For example, ZebraPace, an open-source method for cardiac-rhythm estimation in untethered zebrafish larvae,⁴⁴ or other tools such as HeartBeat software for label-free embryo heartbeat quantification.⁴⁵ These tools apply peak detection and frequency analysis to correct for drift and improve signal robustness. Implementing such methods, along with functional L-DOPA constructs, would enable a more definitive assessment of whether bacteria-derived L-DOPA can meaningfully modulate zebrafish cardiac physiology.

Figure 18: L-DOPA production and effects on zebrafish larval heartbeat. A) Brightfield and mCherry images of a kdrl-negative (i.e., non-fluorescent vasculature) larva, colonized with P. alcaligenes carrying the GB8+7 construct, showing bacterial gut fluorescence (no vascular mCherry). GB7 expresses tyrosine hydroxylase for L-DOPA production. B) Magnified images of the middle gut region, showing bacterial localization. C) Traces of the fluorescence intensity (brightfield signal) over time, showing periodic fluctuations corresponding to cardiac contractions in kdrl:mCherry zebrafish. Fluctuations were extracted from time-lapse recordings where the heart contracts and relaxes under four conditions: control kdrl:mCherry, control kdrl:mCherry treated with 1 mM L-DOPA, Level-1 GB7, and Level-2 GB8+7 strains. D) Quantification of cardiac rhythms in kdrl:mCherry zebrafish under different conditions. Heartbeats per minute (BPM) were extracted from the peak patterns. Control kdrl:mCherry zebrafish averaged to 225 bpm. Control kdrl:mCherry zebrafish supplied with 1mM L-DOPA resulted in 176 bpm, while GB7 (Level 1) and GB8-7 (Level 2) strains produced 278 bpm and 206 bpm, respectively. Data variability and small sample size preclude definitive conclusions.

Discussion

Due to safety regulations, we were forced to keep our bacteria culture at 5℃ in the fridge, meaning TPL was operating at suboptimal conditions. While TPL still had activity, it likely wasn't able to process all the catechol that we were feeding it, resulting in an inhibitory response due to high catechol concentrations.⁴⁶ It is possible that some traces of L-DOPA are present. However, we were unable to use LC-MS/MS even though the protocol could be suitable with minor alterations.

This experiment needs to be re-run under the optimal conditions that were described in our literature protocols. We have run the gels and validated our level 1 constructs previously, therefore it is reasonable to assume that expression of TPL in level 1 plasmid was inducted correctly. The protocol should be conducted at 15℃ with two pumps for automatic, continuous feeding, for 48 hours. This way, it is likely that a new experimental outcome could stand out.

On top of the ex vivo production of L-DOPA experiments, we aimed to determine whether bacterial production of L-DOPA could influence zebrafish cardiac physiology. While minor shifts in heart rate were observed across conditions, the variability between larvae and the presence of a frameshift mutation in the tyrosine phenol lyase cassette make the data inconclusive. The apparent elevation in heartbeat for geneblock 7-colonized larvae and reduction in L-DOPA-treated or geneblocks 8+7 larvae may reflect experimental noise, background fluorescence, or limited sampling rather than true physiological effects. Nevertheless, this work provides an experimental basis for investigating how bacterial L-DOPA synthesis may influence zebrafish physiology in vivo. Future experiments should increase replication of experiments, confirm functional L-DOPA production, and employ automated, validated quantification pipelines such as ZebraPace or HeartBeat to minimize motion artifacts and operator bias. These improvements will clarify whether bacterial-produced L-DOPA can genuinely modulate host cardiovascular dynamics in zebrafish.

Engler, C., & Marillonnet, S. (2014). Golden Gate Cloning. In S. Valla & R. Lale (Eds.), DNA Cloning and Assembly Methods (Vol. 1116, pp. 119–131). Humana Press. https://doi.org/10.1007/978-1-62703-764-8_9
Galloway, N. R., Toutkoushian, H., Nune, M., Bose, N., & Momany, C. (2013). Rapid Cloning For Protein Crystallography Using Type IIS Restriction Enzymes. Crystal Growth & Design, 13(7), 2833–2839. https://doi.org/10.1021/cg400171z
Weber, E., Engler, C., Gruetzner, R., Werner, S., & Marillonnet, S. (2011). A Modular Cloning System for Standardized Assembly of Multigene Constructs. PLoS ONE, 6(2), e16765. https://doi.org/10.1371/journal.pone.0016765
Valenzuela-Ortega, M., & French, C. (2021). Joint universal modular plasmids (JUMP): A flexible vector platform for synthetic biology. Synthetic Biology, 6(1), ysab003. https://doi.org/10.1093/synbio/ysab003
Kostylev, M., Otwell, A. E., Richardson, R. E., & Suzuki, Y. (2015). Cloning Should Be Simple: Escherichia coli DH5α-Mediated Assembly of Multiple DNA Fragments with Short End Homologies. PLOS ONE, 10(9), e0137466. https://doi.org/10.1371/journal.pone.0137466
Xia, H., Chen, H., Cheng, X., Yin, M., Yao, X., Ma, J., Huang, M., Chen, G., & Liu, H. (2022). Zebrafish: An efficient vertebrate model for understanding the role of gut microbiota. Molecular Medicine, 28(1), 161. https://doi.org/10.1186/s10020-022-00579-1
Guzman, L. M., Belin, D., Carson, M. J., & Beckwith, J. (1995). Tight regulation, modulation, and high-level expression by vectors containing the arabinose PBAD promoter. Journal of Bacteriology, 177(14), 4121–4130. https://doi.org/10.1128/jb.177.14.4121-4130.1995
Zhang, Y., Zhang, J., Hara, H., Kato, I., & Inouye, M. (2005). Insights into the mRNA Cleavage Mechanism by MazF, an mRNA Interferase. Journal of Biological Chemistry, 280(5), 3143–3150. https://doi.org/10.1074/jbc.M411811200
iGEM Registry of Standard Biological Parts. Part:BBa_K3036005. Accessed October 3rd 2025. https://parts.igem.org/Part:BBa_K3036005
Soto-Aceves, M. P., Diggle, S. P., & Greenberg, E. P. (2023). Microbial Primer: LuxR-LuxI 5: Quorum Sensing: This article is part of the Microbial Primer collection. Microbiology, 169(9). https://doi.org/10.1099/mic.0.001343
Bhat, E. H., Henard, J. M., Lee, S. A., McHalffey, D., Ravulapati, M. S., Rogers, E. V., Yu, L., Skiles, D., & Henard, C. A. (2024). Construction of a broad-host-range Anderson promoter series and particulate methane monooxygenase promoter variants expand the methanotroph genetic toolbox. Synthetic and Systems Biotechnology, 9(2), 250–258. https://doi.org/10.1016/j.synbio.2024.02.003
iGEM Squirrel Guangzhou Wiki. Parts. Accessed October 3rd 2025. https://2024.igem.wiki/squirrel-guangzhou/
Zhao, A., Sun, J., & Liu, Y. (2023). Understanding bacterial biofilms: From definition to treatment strategies. Frontiers in Cellular and Infection Microbiology, 13, 1137947. https://doi.org/10.3389/fcimb.2023.1137947
iGEM Registry of Standard Biological Parts. Part:BBa_K3036005. Accessed October 3rd 2025. https://parts.igem.org/Part:BBa_K3036005
McMackin, E. A. W., Corley, J. M., Karash, S., Marden, J., Wolfgang, M. C., & Yahr, T. L. (2021). Cautionary Notes on the Use of Arabinose- and Rhamnose-Inducible Expression Vectors in Pseudomonas aeruginosa. Journal of Bacteriology, 203(16). https://doi.org/10.1128/JB.00224-21
Román, R., Lončar, N., Casablancas, A., Fraaije, M. W., & Gonzalez, G. (2020). High-level production of industrially relevant oxidases by a two-stage fed-batch approach: Overcoming catabolite repression in arabinose-inducible Escherichia coli systems. Applied Microbiology and Biotechnology, 104(12), 5337–5345. https://doi.org/10.1007/s00253-020-10622-y
iGEM Registry of Standard Biological Parts. Part:BBa_I0500. Accessed October 6th 2025. https://parts.igem.org/Part:BBa_I0500
Sullivan, C., Matty, M.A., Jurczyszak, D., Gabor, K.A., Millard, P.J., Tobin, D.M., & Kim, C.H. (2017). Infectious disease models in zebrafish. In Methods in Cell Biology (Vol. 138, pp. 101–136). Elsevier. https://doi.org/10.1016/bs.mcb.2016.10.005
Saraceni, P. R., Romero, A., Figueras, A., & Novoa, B. (2016). Establishment of Infection Models in Zebrafish Larvae (Danio rerio) to Study the Pathogenesis of Aeromonas hydrophila. Frontiers in Microbiology, 7. https://doi.org/10.3389/fmicb.2016.01219
Morales, R. A., Rabahi, S., Diaz, O. E., Salloum, Y., Kern, B. C., Westling, M., Luo, X., Parigi, S. M., Monasterio, G., Das, S., Hernández, P. P., & Villablanca, E. J. (2022). Interleukin-10 regulates goblet cell numbers through Notch signaling in the developing zebrafish intestine. Mucosal Immunology, 15(5), 940–951. https://doi.org/10.1038/s41385-022-00546-3
Flores, E. M., Nguyen, A. T., Odem, M. A., Eisenhoffer, G. T., & Krachler, A. M. (2020). The zebrafish as a model for gastrointestinal tract–microbe interactions. Cellular Microbiology, 22(3). https://doi.org/10.1111/cmi.13152
Xia, H., Chen, H., Cheng, X., Yin, M., Yao, X., Ma, J., Huang, M., Chen, G., & Liu, H. (2022). Zebrafish: An efficient vertebrate model for understanding role of gut microbiota. Molecular Medicine, 28(1), 161. https://doi.org/10.1186/s10020-022-00579-1
Ye, L., Bae, M., Cassilly, C. D., Jabba, S. V., Thorpe, D. W., Martin, A. M., Lu, H.-Y., Wang, J., Thompson, J. D., Lickwar, C. R., Poss, K. D., Keating, D. J., Jordt, S.-E., Clardy, J., Liddle, R. A., & Rawls, J. F. (2021). Enteroendocrine cells sense bacterial tryptophan catabolites to activate enteric and vagal neuronal pathways. Cell Host & Microbe, 29(2), 179-196.e9. https://doi.org/10.1016/j.chom.2020.11.011
Smith, T. J., Sundarraman, D., Melancon, E., Desban, L., Parthasarathy, R., & Guillemin, K. (2023). A mucin-regulated adhesin determines the spatial organization and inflammatory character of a bacterial symbiont in the vertebrate gut. Cell Host & Microbe, 31(8), 1371-1385.e6. https://doi.org/10.1016/j.chom.2023.07.003
Lawson, N. D., & Weinstein, B. M. (2002). In Vivo Imaging of Embryonic Vascular Development Using Transgenic Zebrafish. Developmental Biology, 248(2), 307–318. https://doi.org/10.1006/dbio.2002.0711
Proulx, K., Lu, A., & Sumanas, S. (2010). Cranial vasculature in zebrafish forms by angioblast cluster-derived angiogenesis. Developmental Biology, 348(1), 34–46. https://doi.org/10.1016/j.ydbio.2010.08.036
Shinoda, H., Shannon, M., & Nagai, T. (2018). Fluorescent Proteins for Investigating Biological Events in Acidic Environments. International Journal of Molecular Sciences, 19(6), 1548. https://doi.org/10.3390/ijms19061548
Richard, J. P., Melikov, K., Brooks, H., Prevot, P., Lebleu, B., & Chernomordik, L. V. (2005). Cellular uptake of unconjugated TAT peptide involves clathrin-dependent endocytosis and heparan sulfate receptors. The Journal of biological chemistry, 280(15), 15300–15306. https://doi.org/10.1074/jbc.M401604200
Tünnemann, G., Martin, R. M., Haupt, S., Patsch, C., Edenhofer, F., & Cardoso, M. C. (2006). Cargo-dependent mode of uptake and bioavailability of TAT-containing proteins and peptides in living cells. FASEB journal, 20(11), 1775–1784. https://doi.org/10.1096/fj.05-5523com
Verweij, F. J., Revenu, C., Arras, G., Dingli, F., Loew, D., Pegtel, D. M., Follain, G., Allio, G., Goetz, J. G., Zimmermann, P., Herbomel, P., Del Bene, F., Raposo, G., & Van Niel, G. (2019). Live Tracking of Inter-organ Communication by Endogenous Exosomes In Vivo. Developmental Cell, 48(4), 573-589.e4. https://doi.org/10.1016/j.devcel.2019.01.004
Suzuki, M., Suzuki, S., Matsui, M., Hiraki, Y., Kawano, F., & Shibayama, K. (2013). Genome Sequence of a Strain of the Human Pathogenic Bacterium Pseudomonas alcaligenes That Caused Bloodstream Infection. Genome announcements, 1(5), e00919-13. https://doi.org/10.1128/genomeA.00919-13
James, S., Reddy, V., & Beato, M.R. (2024). National Library of Medicine. Neuroanatomy, Substantia Nigra. Accessed October 3rd 2025. https://www.ncbi.nlm.nih.gov/books/NBK536995/
Triarhou, L. National Library of Medicine. Dopamine and Parkinson's Disease. Accessed October 3rd 2025. https://www.ncbi.nlm.nih.gov/books/NBK6271/
Salat, D., & Tolosa, E. (2013). Levodopa in the Treatment of Parkinson's Disease: Current Status and New Developments. Journal of Parkinson's Disease, 3(3), 255–269. https://doi.org/10.3233/JPD-130186
Beckers, M., Bloem, B. R., & Verbeek, M. M. (2022). Mechanisms of peripheral levodopa resistance in Parkinson's disease. Npj Parkinson's Disease, 8(1), 56. https://doi.org/10.1038/s41531-022-00321-y
Noack, C., Schroeder, C., Heusser, K., & Lipp, A. (2014). Cardiovascular effects of levodopa in Parkinson's disease. Parkinsonism & Related Disorders, 20(8), 815–818. https://doi.org/10.1016/j.parkreldis.2014.04.007
Stednitz, S. J., Freshner, B., Shelton, S., Shen, T., Black, D., & Gahtan, E. (2015). Selective toxicity of L-DOPA to dopamine transporter-expressing neurons and locomotor behavior in zebrafish larvae. Neurotoxicology and Teratology, 52, 51–56. https://doi.org/10.1016/j.ntt.2015.11.001
Kwon, D. K., Kwatra, M., Wang, J., & Ko, H. S. (2022). Levodopa-Induced Dyskinesia in Parkinson's Disease: Pathogenesis and Emerging Treatment Strategies. Cells, 11(23), 3736. https://doi.org/10.3390/cells11233736
Kumagai, H., Katayama, T., Koyanagi, T., & Suzuki, H. (2023). Research overview of L-DOPA production using a bacterial enzyme, tyrosine phenol-lyase. Proceedings of the Japan Academy Series B, 99(3), 75–101. https://doi.org/10.2183/pjab.99.006
Fordjour, E., Adipah, F. K., Zhou, S., Du, G., & Zhou, J. (2019). Metabolic engineering of Escherichia coli BL21 (DE3) for de novo production of l-DOPA from d-glucose. Microbial Cell Factories, 18(1). https://doi.org/10.1186/s12934-019-1085-z
Xu, D.-Y., Chen, J.-Y., & Yang, Z. (2012). Use of cross-linked tyrosinase aggregates as catalyst for synthesis of L-DOPA. Biochemical Engineering Journal, 63, 88-94. https://doi.org/10.1016/j.bej.2011.11.009
Gore, M., & Burggren, W. W. (2012). Cardiac and Metabolic Physiology of Early Larval Zebrafish (Danio rerio) Reflects Parental Swimming Stamina. Frontiers in Physiology, 3. https://doi.org/10.3389/fphys.2012.00035
Verweij, F. J., Revenu, C., Arras, G., Dingli, F., Loew, D., Pegtel, D. M., Follain, G., Allio, G., Goetz, J. G., Zimmermann, P., Herbomel, P., Del Bene, F., Raposo, G., & Van Niel, G. (2019). Live Tracking of Inter-organ Communication by Endogenous Exosomes In Vivo. Developmental Cell, 48(4), 573-589.e4. https://doi.org/10.1016/j.devcel.2019.01.004
Gaur, H., Pullaguri, N., Nema, S., Purushothaman, S., Bhargava, Y., & Bhargava, A. (2018). ZebraPace: An Open-Source Method for Cardiac-Rhythm Estimation in Untethered Zebrafish Larvae. Zebrafish, 15(3), 254–262. https://doi.org/10.1089/zeb.2017.1545
Gierten, J., Pylatiuk, C., Hammouda, O. T., Schock, C., Stegmaier, J., Wittbrodt, J., Gehrig, J., & Loosli, F. (2020). Automated high-throughput heartbeat quantification in medaka and zebrafish embryos under physiological conditions. Scientific Reports, 10(1). https://doi.org/10.1038/s41598-020-58563-w
Han, H., Zeng, W., Du, G., Chen, J., & Zhou, J. (2020). Site-directed mutagenesis to improve the thermostability of tyrosine phenol-lyase. Journal of Biotechnology, 310, 6–12. https://doi.org/10.1016/j.jbiotec.2020.01.005

Results

Overview

Engineered Function

FPs in the Gut

Transwell System

L-DOPA Production

Engineered Function

Introduction

Background

GGA and the pJUMP system

Bacterial hosts

Kill switches

Aim

Results

Validating and assessing the constructs

Testing the functionality of the L-arabinose kill switch

Assessing the functionality of fuGFP secretion

Conclusion and Outlooks

FPs in the Gut

Introduction

Background

Zebrafish as a model organism

Organizational and functional aspects of the zebrafish gut

Transgenic zebrafish lines used in this study

Aim

Results

Engineered bacteria stably colonize the zebrafish larval gut

Engineered bacteria colonize multiple areas of the zebrafish larval gut, including those where nutrient uptake is happening

Engineered bacteria show differences in uptake

In vivo activation of an arabinose-induced kill switch in gut colonizing bacteria

Conclusion

Transwell System

Introduction

Experiment 1: Transwell Testing with Added Compounds

Results and Discussion

Table 1: Added Working Solution

Table 2: TEER Results at t=0 and t=4, After a 4 Hour Incubation Period

Discussion

Experiment 2: Toxicity Assay

Table 3: Layout of Medium Samples Collected from the Toxicity Assay

Table 4: LDH Release Measurements in Caco-2 Toxicity Assay

Discussion

Experiment 3: Bacterial Infection

Results

Discussion

Experiment 4: GFP Measurement

Discussion

L-DOPA Production

Introduction

Background

L-DOPA and the physiological response

Ex vivo bacteria-based production of L-DOPA

L-DOPA Production

Production Design

Aim

Results

Table 5: Standard Curve Results

L-DOPA production and effects on zebrafish larval heartbeat

Discussion

References